1 | package ocraction;
|
---|
2 |
|
---|
3 | use strict;
|
---|
4 |
|
---|
5 | use cgiactions::baseaction;
|
---|
6 |
|
---|
7 | use dbutil;
|
---|
8 | use ghtml;
|
---|
9 |
|
---|
10 |
|
---|
11 | BEGIN {
|
---|
12 | # unshift (@INC, "$ENV{'GSDLHOME'}/perllib/cpan/perl-5.8");
|
---|
13 | require XML::Rules;
|
---|
14 | }
|
---|
15 |
|
---|
16 |
|
---|
17 | @ocraction::ISA = ('baseaction');
|
---|
18 |
|
---|
19 | my $action_table =
|
---|
20 | {
|
---|
21 | "preformOcr" => { 'compulsory-args' => [ "d", "imagename" ],
|
---|
22 | 'optional-args' => [] },
|
---|
23 | "setText" => { 'compulsory-args' => [ "d", "newtext" ],
|
---|
24 | 'optional-args' => [] }
|
---|
25 | };
|
---|
26 |
|
---|
27 | sub new
|
---|
28 | {
|
---|
29 | my $class = shift (@_);
|
---|
30 | my ($gsdl_cgi,$iis6_mode) = @_;
|
---|
31 |
|
---|
32 | my $self = new baseaction($action_table,$gsdl_cgi,$iis6_mode);
|
---|
33 |
|
---|
34 | return bless $self, $class;
|
---|
35 | }
|
---|
36 |
|
---|
37 | sub preformOcr
|
---|
38 | {
|
---|
39 | my $self = shift @_;
|
---|
40 |
|
---|
41 | my $collection = $self->{'collect'};
|
---|
42 | my $site = $self->{'site'};
|
---|
43 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
---|
44 |
|
---|
45 | my $docId = $self->{'d'};
|
---|
46 | my $imagename = $self->{'imagename'};
|
---|
47 |
|
---|
48 | # Change URL style '/' to Windows style
|
---|
49 | $imagename =~ s/\//\\/g;
|
---|
50 |
|
---|
51 | my $indexDir = "C:\\Users\\Bryce\\Desktop\\Research\\greenstone3-svn-bryce64\\web\\sites\\localsite\\collect\\jonesmin\\";
|
---|
52 | ##my $hashDir = $archivesDir . substr($docId,0,8) . ".dir\\";
|
---|
53 |
|
---|
54 | my $imageFile = $indexDir . $imagename;
|
---|
55 |
|
---|
56 | my $outputFile = $ENV{'TEMP'} . "\\out";
|
---|
57 | my $cmd = "tesseract $imageFile $outputFile";
|
---|
58 | ## print STDERR "\n\n CMD=\n$cmd \n\n";
|
---|
59 |
|
---|
60 | my $status = system($cmd);
|
---|
61 | if($status != 0) { print STDERR "\n\n Fail to run \n $cmd \n $! \n\n"; }
|
---|
62 | if (open (FILE, "<$outputFile" . ".txt")==0) {
|
---|
63 | $gsdl_cgi->generate_error("Unable to open file containing OCR'd text:<br> $!");
|
---|
64 | return;
|
---|
65 | }
|
---|
66 |
|
---|
67 | my $result = "";
|
---|
68 | my $line;
|
---|
69 | while(defined ($line=<FILE>)) {
|
---|
70 | $result .= $line;
|
---|
71 | }
|
---|
72 | close FILE;
|
---|
73 | unlink($outputFile . ".txt");
|
---|
74 |
|
---|
75 | $gsdl_cgi->generate_ok_message($result);
|
---|
76 | }
|
---|
77 |
|
---|
78 | sub setText
|
---|
79 | {
|
---|
80 | my $self = shift @_;
|
---|
81 | my $collection = $self->{'collect'};
|
---|
82 | my $site = $self->{'site'};
|
---|
83 | my $docId = $self->{'d'};
|
---|
84 | my $newtext = $self->{'newtext'};
|
---|
85 | }
|
---|
86 |
|
---|
87 | 1; |
---|